Skip Navigation


Journal of Petrology Advance Access originally published online on June 6, 2006
Journal of Petrology 2006 47(10):1915-1942; doi:10.1093/petrology/egl031
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
47/10/1915    most recent
egl031v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (11)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by MARSCHALL, H. R.
Right arrow Articles by TONARINI, S.
Right arrow Search for Related Content
GeoRef
Right arrow GeoRef Citation
Social Bookmarking
 Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© The Author 2006. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oxfordjournals.org

Syros Metasomatic Tourmaline: Evidence for Very High-{delta}11B Fluids in Subduction Zones

HORST R. MARSCHALL1,*, THOMAS LUDWIG1, RAINER ALTHERR1, ANGELIKA KALT2 and SONIA TONARINI3

1 MINERALOGISCHES INSTITUT, UNIVERSITÄT HEIDELBERG IM NEUENHEIMER FELD 236, D-69120 HEIDELBERG, GERMANY
2 INSTITUT DE GÉOLOGIE, UNIVERSITÉ DE NEUCHÂTEL RUE EMILE ARGAND 11, CH-2007 NEUCHÂTEL, SWITZERLAND
3 ISTITUTO DI GEOSCIENZE E GEORISORSE, CNR VIA MORUZZI 1, I-56124 PISA, ITALY

RECEIVED SEPTEMBER 19, 2005; ACCEPTED MAY 8, 2006


    ABSTRACT
 TOP
 ABSTRACT
 INTRODUCTION
 ANALYTICAL METHODS
 GEOLOGICAL BACKGROUND
 INVESTIGATED SAMPLES
 RESULTS
 DISCUSSION
 SUMMARY AND CONCLUSIONS
 APPENDIX: EQUATIONS USED FOR...
 REFERENCES
 
High-pressure (HP) metamorphic blocks enclosed in a mafic to ultramafic matrix from a mélange on the island of Syros are rimmed by tourmaline-bearing reaction zones (blackwalls). The B isotopic composition of dravitic tourmaline within these blackwalls was investigated in situ by secondary ion mass spectrometry. Boron in these tourmalines is unusually heavy, with {delta}11B values exceeding +18{per thousand} in all investigated samples and reaching an extreme value of +28·4{per thousand} in one sample. Blackwalls formed during exhumation of the HP mélange at a depth of 20–25 km at temperatures of 400–430°C, by influx of external hydrous fluids. The compositions of the fluids are estimated to be in the range of 100–300 µg/g B with {delta}11B values of +18 to +28{per thousand}. The high {delta}11B values cannot be explained by tourmaline formation from unmodified slab-derived fluids. However, such fluids could interact with the material in the exhumation channel on their way from the dehydrating slab to the site of tourmaline formation in the blackwalls. This could produce exceptionally high {delta}11B values in the fluids, a case that is modelled in this study. The model demonstrates that subduction fluids may be effectively modified in both trace element and isotopic composition during their migration through the material overlying the subducting slab. Blackwall tourmaline from Syros has a large grain size (several centimetres), high abundance, and an exceptionally high {delta}11B value. The formation of tourmaline at the contact between mafic or felsic HP blocks and their ultramafic matrix involved fluids released during dehydration reactions in the subducting slab. It forms a heavy-boron reservoir in hybrid rocks overlying the subducting slab, and may, thus, have a significant impact on the geochemical cycle of B and its isotopes in subduction zones.

KEY WORDS: boron isotopes; tourmaline; subduction zone; fluid, high pressure


    INTRODUCTION
 TOP
 ABSTRACT
 INTRODUCTION
 ANALYTICAL METHODS
 GEOLOGICAL BACKGROUND
 INVESTIGATED SAMPLES
 RESULTS
 DISCUSSION
 SUMMARY AND CONCLUSIONS
 APPENDIX: EQUATIONS USED FOR...
 REFERENCES
 
Important constraints on the mechanisms of slab-to-mantle material transfer can be derived from the abundance and isotopic systematics of boron in subduction-related rocks. Element ratios such as B/Be and B/Nb and B isotope signatures have been used over the last ~20 years to enhance our knowledge of slab dehydration and subduction-related melting processes. The high relative mass difference of the two stable isotopes of boron (10B and 11B) is responsible for variations in its isotopic composition in nature of about 100{per thousand} (Palmer & Swihart, 2002Go). The large B isotopic fractionation at low temperatures has led to the use of the B stable isotope system for studies of processes acting at the Earth's surface (e.g. Hemming & Hanson, 1992Go; Barth, 1998Go) and for the detection of chemical and isotopic signals of subducted material in fluids, serpentinites and subduction-related volcanic rocks (e.g. Ishikawa & Nakamura, 1994Go; Benton et al., 2001Go). B is known to be relatively mobile in hydrous fluids and in silicate melts (e.g. Brenan et al., 1998aGo, 1998bGo). B concentrations in the slab decrease with increasing depth of subduction, as it is released with fluids during dehydration (Moran et al., 1992Go; Bebout et al., 1999Go). Concurrently, preferential loss of the heavier isotope 11B results in decreasing {delta}11B values in the subducting material (Peacock & Hervig, 1999Go). Studies on subduction-related volcanic rocks investigating across-arc transects show a decrease of B concentrations, B/Be and B/Nb ratios and {delta}11B values from trench to back-arc (Ishikawa & Nakamura, 1994Go; Ishikawa & Tera, 1997Go; Benton et al., 2001Go; Ishikawa et al., 2001Go). Within the slab, the most important reservoirs for B are sediments, altered magmatic rocks and serpentinized ultramafic rocks (e.g. Leeman & Sisson, 2002Go; Scambelluri et al., 2004Go). Fresh mantle and magmatic rocks have very low B concentrations and a relatively uniform B isotope ratio. Sediments, altered magmatic rocks and serpentinites are not only enriched in B, but also show {delta}11B values higher than mid-ocean ridge basalt (MORB) (or mantle), as a result of the interaction with seawater, which has a very high {delta}11B value. The hydrous portion of the subducting slab, therefore, introduces large amounts of heavy B into the subduction zone.

Tourmaline
Tourmaline is the most widespread borosilicate mineral in natural rocks. It contains ~3 wt % B, an element that is of very low abundance in most crustal and mantle rocks (e.g. Ryan & Langmuir, 1993Go; Chaussidon & Jambon, 1994Go; McDonough & Sun, 1995Go; Taylor & McLennan, 1995Go). Therefore, formation of tourmaline requires effective concentration of B, even if it is formed in rocks of the continental crust. This concentration is attained either (1) by magmatic differentiation, forming tourmaline in pegmatites enriched in B as a result of its highly incompatible character, or (2) by fluid-dominated processes because of the highly fluid-mobile character of B. The B necessary for tourmaline growth in fluid-saturated rocks may be supplied by other phases within the rock itself, such as clay minerals or mica (Nakano & Nakamura, 2001Go) or by external fluids during metasomatism (Altherr et al., 2004Go). Once formed, tourmaline is highly stable in a large variety of rock types over an exceptionally large PT range, including conditions in subducting slabs to depths of ~250 km (Werding & Schreyer, 2002Go).

Tourmaline from (ultra-)HP metamorphic rocks has been used to unravel the evolution of the B isotopic composition of rocks composing the subducted slab. It has been demonstrated by Bebout & Nakamura (2003)Go that the {delta}11B values of tourmaline decrease during prograde growth, are very low in HP growth zones, and increase in retrograde growth zones. Intracrystalline isotopic equilibration by diffusion is very limited and different growth zones still display their initial isotopic compositions. Tourmaline, therefore, provides a powerful tool to monitor the B isotope evolution of metamorphic rocks and provides important insights into the B isotope budget of metamorphic fluids.

Boron isotope ratios of tourmaline from metasomatic reaction zones in HP metamorphic rocks from the island of Syros are presented in this paper. Implications for {delta}11B values of subduction zone fluids, together with a model providing an explanation for the extremely heavy B isotopic composition of Syros tourmaline, will be discussed.


    ANALYTICAL METHODS
 TOP
 ABSTRACT
 INTRODUCTION
 ANALYTICAL METHODS
 GEOLOGICAL BACKGROUND
 INVESTIGATED SAMPLES
 RESULTS
 DISCUSSION
 SUMMARY AND CONCLUSIONS
 APPENDIX: EQUATIONS USED FOR...
 REFERENCES
 
Compositions of mineral phases were determined using a Cameca SX 51 electron microprobe equipped with five wavelength-dispersive spectrometers (Mineralogisches Institut, Heidelberg). Operating conditions were 20 nA beam current and 15 kV acceleration voltage. For analyses of phengite, the electron beam was defocused to 10 µm to avoid loss of alkalis. For analyses of tourmaline it was defocused to 5 µm. Details on counting times, crystals, standards and detection limits have been given by Marschall (2005)Go. PAP correction was applied to the raw data (Pouchou & Pichoir, 1984Go, 1985Go).

For tourmaline, a modified matrix correction was applied assuming stoichiometric oxygen and all non-measured components to be B2O3. The accuracy of the electron microprobe analyses of tourmaline and the correction procedure was checked by measuring three samples of reference tourmalines (98144, elbaite; 108796, dravite; 112566, schorl; Dyar et al., 1998Go, 2001Go). Under the described conditions, analytical errors on all analyses are ±1% relative for major elements and ±5% relative for minor elements. A detailed description of the electron microprobe techniques for tourmaline analysis has been given by Kalt et al. (2001)Go.

Concentrations of H, Li and B and B isotope ratios were measured by secondary ion mass spectrometry (SIMS) with a modified Cameca IMS 3f ion microprobe at the Mineralogisches Institut, Heidelberg, equipped with a primary beam mass filter. Secondary ions 7Li and 11B in tourmaline were collected under an ion-imaged field of 150 µm diameter (Kalt et al., 2001Go). Secondary ions of 1H were collected by reducing the 25 µm beam spot to an ion-imaged field of 12 µm by using a field aperture, to minimize the influence of surface contamination (Marschall & Ludwig, 2004Go). Water contamination was further reduced using a cold-trap cooled with liquid nitrogen attached to the sample chamber of the IMS 3f. The relative ion yields (RIY) for H and B were determined using three tourmalines as reference material: elbaite (98144), dravite (108796) and schorl (112566). The relative reproducibility (1{sigma}) of the analyses of B was <1%. For Li the reference material was NIST SRM 610 (Pearce et al., 1997Go) with a relative reproducibility of <1%. The accuracy is limited by matrix effects and the uncertainty of the element concentrations in the reference material; the relative uncertainty is estimated to be <25% for H, <20% for Li and <10% for B (Kalt et al., 2001Go). Between 10 and 20 Li and B analyses and 5–10 H analyses were performed on tourmaline in each sample.

The analysis of boron isotope ratios required significant changes of both software and hardware of the SIMS. Most important were changes in the electronic control device of the magnet, which permitted faster switches between different masses and better stability of the magnet. The changes have been described in detail by Marschall (2005).Go

The primary ion beam was 16O accelerated to 10 keV with a beam current of 1 nA, resulting in count rates for 11B of ~2 x 105 s–1 and ~5 x 104 s–1 for 10B on tourmaline, collected by a single electron multiplier. The diameter of the 1 nA spot was ≤5 µm. The energy window was set to 50 eV and no offset was applied. Fifty cycles were measured on each analysis spot with counting times of 3·307 s and 1·660 s on 10B and 11B, respectively. Presputtering lasted for 5 min and the settling time between two different masses was 200 ms, resulting in total analysis time for one spot of ~10 min. The internal precision of a single analysis was ≤1{per thousand} (2{sigma}). Boron isotopic compositions of samples are reported using the {delta}-notation ({delta}11B in {per thousand}) relative to the SRM 951 accepted value (Catanzaro et al., 1970Go). Instrumental mass fractionation was corrected for by using three samples of reference tourmaline (98114, elbaite; 108796, dravite; 112566, schorl; Leeman & Tonarini, 2001Go). Correction factors {alpha}inst for the different reference tourmalines resulted in a mean {alpha}inst of 1·0476. Reproducibility of measured isotope ratios during the analytical session (8 days) was ±0·5{per thousand}.

During analysis of tourmaline samples by SIMS, traverses through different grains were measured to check for possible intra-grain variations in {delta}11B. Between 10 and 50 boron isotope analyses were performed for each sample. SIMS analysis of our samples revealed very high 11B/10B ratios in contrast to the reference tourmalines, which range from –12·5{per thousand} to –6·6{per thousand} (Leeman & Tonarini, 2001Go). To ensure consistency, three tourmaline samples from Syros were analysed by positive thermal ion mass spectrometry (P-TIMS), following the method of Tonarini et al. (1997)Go. Correlation between SIMS and P-TIMS results for three Syros samples and the three reference tourmalines demonstrate good correlation between the two methods (see Fig. 2) used in the two laboratories. This proves the instrumental mass fractionation of SIMS to be independent of 11B/10B ratios in the range of {delta}11B in natural tourmaline.


    GEOLOGICAL BACKGROUND
 TOP
 ABSTRACT
 INTRODUCTION
 ANALYTICAL METHODS
 GEOLOGICAL BACKGROUND
 INVESTIGATED SAMPLES
 RESULTS
 DISCUSSION
 SUMMARY AND CONCLUSIONS
 APPENDIX: EQUATIONS USED FOR...
 REFERENCES
 
The island of Syros displays a sequence of rocks belonging to the lower unit of the Attic–Cycladic Crystalline Complex (Dürr et al., 1978Go). The major part of the island is composed of interlayered schists and marbles dipping north to NE (e.g. Hecht, 1984Go; Dixon & Ridley, 1987Go; Seck et al., 1996Go). They were interpreted as metamorphosed flysch sediments by Dixon & Ridley (1987)Go. The most interesting formations are exposed in the northern part of Syros near Kámpos and along the coastline around Hermoupolis and Kini (Fig. 1). These formations are composed of metagabbro, eclogite, glaucophane schist, meta-plagiogranite, serpentinite and metasediment, which all mainly preserve a blueschist- to eclogite-facies metamorphic overprint. The structural and tectono-metamorphic evolution of the island of Syros and especially its northern part has been investigated in a large number of studies (Dixon, 1968Go; Bonneau et al., 1980Go; Ridley, 1982Go, 1984Go, 1986Go; Ballhaus & Schumacher, 1995Go; Trotet et al., 2001aGo; Rosenbaum et al., 2002Go; Ring et al., 2003Go; Brady et al., 2004Go; Keiter et al., 2004Go). Most workers distinguish three ductile deformation phases along a single clockwise PT path: (1) prograde deformation; (2) deformation near the peak pressure of metamorphism; (3) retrograde deformation. The prograde PT path is characterized by a high P/T ratio, typical for subduction zone metamorphism. Peak metamorphic conditions have been estimated at ~470–520°C and 1·3–2·0 GPa by different workers for different rock types (Dixon, 1968Go; Ridley, 1984Go; Okrusch & Bröcker, 1990Go; Trotet et al., 2001bGo; Rosenbaum et al., 2002Go; Keiter et al., 2004Go). The exhumation path of the rocks is characterized by near-isothermal decompression, resulting in the preservation of the HP assemblages and minerals. Trotet et al. (2001bGo) showed that the rocks from Syros were decompressed at temperatures between 400 and 550°C to pressures between 1·0 and 0·5 GPa. For a clinopyroxene + albite + glaucophane-bearing retrograde assemblage they calculated conditions of 0·78 GPa and 394°C. This is in line with the occurrence of fresh lawsonite in some places around the island (Dixon, 1968Go; Brady et al., 2001Go; Marschall, 2005Go), because this mineral easily decomposes upon heating.


Figure 1
View larger version (29K):
[in this window]
[in a new window]
[Download PowerPoint slide]
 
Fig. 1 Simplified geological map of Syros after Hecht (1984)Go, showing sample localities within the mafic–ultramafic mélange near Kampos. Inset shows a map of Greece with the location of the island of Syros.

 


Figure 2
View larger version (15K):
[in this window]
[in a new window]
[Download PowerPoint slide]
 
Fig. 2 Comparison of B isotope ratios of tourmaline determined by SIMS and TIMS, expressed in the {delta}-notation relative to NIST SRM 951 (Catanzaro et al., 1970Go). {circ}, Syros samples; •, standard tourmalines (Leeman & Tonarini, 2001Go). For sample SY400, core and rim values differ by ~1·3{per thousand}, which is shown by the two data points. Error bars show 2{sigma} errors of SIMS (~1{per thousand}). Errors for TIMS (~0·5{per thousand}) are within the size of the symbols.

 
The above-mentioned units of metabasites occur in the upper level of the major marble–schist sequence. The internal structure of the northern mélange in the area of Kámpos (Fig. 1), which is the best investigated area, is rather complex. A number of rock types have been distinguished on the basis of geochemical and petrographic observations within an east–west-trending zone. They are interpreted as the high-pressure metamorphic equivalents of different parts of ancient oceanic crust (e.g. Seck et al., 1996Go). West of Kámpos, the various rock types form a mélange with blocks of eclogite, omphacite/jadeite fels (the term fels in this paper refers to a metamorphic rock with an isotropic fabric), garnet glaucophane fels, serpentinite and metagabbro embedded in a matrix of chlorite schist and serpentinite (Dixon, 1968Go; Hecht, 1984Go; Dixon & Ridley, 1987Go; Okrusch & Bröcker, 1990Go; Seck et al., 1996Go; Bröcker & Enders, 2001Go). Contact between the blocks and their chemically and mineralogically contrasting matrix are characterized by reaction zones rich in OH-bearing minerals (Fig. 3a), which are referred to subsequently as ‘blackwalls’. Blackwall rocks are hybrids as they have no pre-metamorphic equivalent of comparable chemical composition, but were formed from two different precursor rocks by tectonic mixing and/or metasomatic transfer of material. This process of blackwall formation on Syros was previously described by Dixon (1968)Go and other workers (Ridley, 1984Go; Dixon & Ridley, 1987Go; Okrusch & Bröcker, 1990Go; Ballhaus & Schumacher, 1995Go; Seck et al., 1996Go; Bröcker & Enders, 2001Go), and is also known from other high-pressure terranes, such as Cima di Gagnone, Alps, Switzerland (Rice et al., 1974Go; Evans et al., 1979Go), Santa Catalina Island, California, USA (Sorensen & Grossman, 1989Go; Bebout & Barton, 2002Go), the Franciscan Complex, California, USA (King et al., 2003Go), Port Macquarie, New South Wales, Australia (Och et al., 2003Go) and NE New Caledonia (Carson et al., 2000Go; Spandler et al., 2003Go; Fitzherbert et al., 2004Go).


Figure 3
View larger version (116K):
[in this window]
[in a new window]
[Download PowerPoint slide]
 
Fig. 3 Different varieties of blackwalls in the field, in hand specimen and in thin section. (a) Omphacite (Omp) fels enveloped by an ~10 cm thick layer of almost pure chlorite (Chl). (b) Tourmaline (Tur, black) covering an HP block in a dense layer. (c) Tourmaline and apatite (Ap) at the contact between an omphacite fels and the surrounding glaucophane schist. (d) Centimetre-sized tourmaline crystals in the outer zone of an omphacite fels. (e) Plane-polarized light photomicrograph of thin section of sample SY412, showing an idioblast of garnet (Grt), truncated by a chlorite-filled crack and enclosed in a large tourmaline crystal (dark). Omphacite (Omp) included in tourmaline also shows type-I cores partly replaced by type-II rims, similar to the omphacite in the matrix. (f) Photomicrograph of thin section of sample SY441, showing large zoned tourmaline crystals in paragenesis with chlorite (unpolarized light). (g) and (h) are photomicrographs of thin section of sample SY442. (g) Zoned tourmaline crystal with small omphacite inclusions. The thin section was covered with a gold layer prior to SIMS analysis, producing the dark blue colours of the picture. The spots generated during SIMS analyses are visible as white dots. I, II, III and IV refer to zones I–IV, respectively (see text and Fig. 7). (h) Detail of the matrix surrounding tourmaline in (g) consisting of euhedral and subhedral omphacite intergrown with albite.

 

    INVESTIGATED SAMPLES
 TOP
 ABSTRACT
 INTRODUCTION
 ANALYTICAL METHODS
 GEOLOGICAL BACKGROUND
 INVESTIGATED SAMPLES
 RESULTS
 DISCUSSION
 SUMMARY AND CONCLUSIONS
 APPENDIX: EQUATIONS USED FOR...
 REFERENCES
 
The eight blackwall samples investigated in this study (Table 1) were taken from different places within the mélange (Fig. 1). They all contain significant amounts of tourmaline, concentrated in millimetre- to centimetre-thick layers on the surface of the high-pressure blocks (Fig. 3b and c), or more evenly distributed single crystals or clusters of crystals (Fig. 3d) within them. Two samples (SY309B, SY420) are schists from the matrix, which contain tourmaline-rich layers or veins. Apart from tourmaline, the blackwall rocks contain omphacite (Jd35–53), chlorite (clinochlore), phengite (3·3–3·4 Si per 11 oxygens), and almost pure albite (Ab97–100). Some samples additionally contain biotite (Table 1). Quartz was not observed in any of the samples. Tourmaline in all samples is euhedral and does not show any sign of brittle or ductile deformation. In thin section, the grains commonly show visible zonation. Colours of the pleochroic cores change from colourless to blue, whereas those of the rims change from pale pink to green. Rims of tourmaline in some samples (SY400, SY441, SY442) show two or three zones with different shades of green. Resorption of core domains and replacement by rim material is limited, and resorption of rim domains was never observed.


View this table:
[in this window]
[in a new window]

 
Table 1 List of tourmaline-bearing blackwall and mélange matrix samples investigated for B isotope ratios

 
Sample SY309 is composed of glaucophane schist (SY309A) and tourmaline-rich layers (SY309B). The latter are concordant with the foliation of the host rock and can be tracked over several metres in the outcrop. Sample SY309B, a 1–2 cm thick layer consisting of tourmaline + omphacite + chlorite + albite + phengite + glaucophane + epidote + apatite + rutile + titanite, is hosted in a glaucophane schist composed of glaucophane + epidote + phengite + titanite. Tourmaline crystals are 0·1–1·0 mm in diameter and 1–6 mm in length with some inclusions of titanite, rutile, omphacite, albite (Ab99·7) and apatite.

Sample SY327 is a dark green omphacite–chlorite fels consisting of equal amounts of chlorite and omphacite, making up about 80% of the rock. Additional phases are abundant apatite (2–3 mm), albite, ilmenite, titanite, biotite, amphibole and tourmaline. Three compositional types of clinopyroxene were observed in the matrix. The first type forms the cores of large omphacite grains (1–5 mm) with intermediate acmite and jadeite contents (Jd37Acm28Q34). The second type is found as rims of the same grains with a much higher jadeite content (Jd52Acm27Q19). The third type is found in patches composed of small grains (10 µm) closely intergrown with albite (Ab97). Type-3 clinopyroxene has high acmite and very low jadeite contents (Jd15Acm49Q35 = aegirine). Garnet is not preserved in the rock, but was probably present before, as indicated by hexagonal pseudomorphs consisting of fine-grained chlorite. Amphibole occurs only as small inclusions (<100 µm) in tourmaline and is zoned from blue in the core to bluish green at the rim, with compositions changing from glaucophane to nyböite and magnesiokataphorite. Tourmaline grains are ~4 mm in diameter and 3 cm in length and are highly poikiloblastic with inclusions of predominantly omphacite, blue–green amphibole, apatite and chlorite.

Sample SY400 is a pale green tourmaline–omphacite–chlorite fels (Fig. 3d), consisting primarily of chlorite and fine-grained omphacite with minor epidote, phengite and accessory rutile, titanite and apatite. Tourmaline crystals are 2–5 mm in diameter and 10–20 mm in length and highly poikiloblastic with inclusions of omphacite, rutile, titanite, apatite and epidote. The rock itself shows healed cracks (about 1 mm wide) filled by chlorite and prismatic omphacite, grown perpendicular to the cracks' elongation. The cracks can be traced pseudomorphically through the large tourmaline grains by zones of lower inclusion density. Tourmaline itself is not deformed or cracked. Therefore, tourmaline growth obviously postdates both chloritization and deformation.

Sample SY412 was taken from an interior part of an eclogite block, which shows a transition into a glaucophane-rich assemblage at its margins (see SY413 below). The eclogitic paragenesis of SY412 is garnet + type-1 omphacite + rutile + phengite + glaucophane + epidote (± quartz?). This primary assemblage is partly replaced by a secondary assemblage of dominantly hydrous minerals, which form small, closely intergrown aggregates between the larger primary phases. The secondary assemblage is composed of phengite + glaucophane + albite + chlorite + type-2 omphacite + tourmaline. In some patches, biotite formed at the expense of phengite. The eclogitic garnet contains inclusions of glaucophane and is partly pseudomorphosed by chlorite. It shows chemical zonation with decreasing Mn content and increasing Fe and Mg contents from core to rim (Table 2). Omphacite is chemically zoned with type-1 cores (Jd42Acm15Q41) and type-2 omphacite rims (Jd53Acm30Q14). Rutile is still abundant and shows only thin overgrowths of titanite (<5 µm). Tourmaline forms large grains 1–3 mm in diameter and 5–15 mm in length, with inclusions of omphacite, glaucophane, garnet, phengite, epidote, albite and rutile (Fig. 3e). Garnet grains included in tourmaline show incipient chloritization along cracks, which was obviously stopped by the shielding tourmaline (Fig. 3e). Furthermore, tourmaline locally grew into the chlorite-filled cracks in garnet. Therefore, we conclude that tourmaline replaced chlorite, but did not interact with garnet. Garnet rims are in direct contact with tourmaline and type-2 omphacite. The observed petrographic relationship of the minerals (i.e. tourmaline, omphacite, chlorite and garnet) are very similar to those described by Altherr et al. (2004)Go.


View this table:
[in this window]
[in a new window]

 
Table 2 Representative electron microprobe analyses of different minerals

 
Sample SY413 is part of the glaucophane-rich envelope of the above-described block of sample SY412. The rock is dominated by glaucophane and chlorite pseudomorphs after garnet, accompanied by phengite, omphacite, albite, titanite, tourmaline and epidote. Long-prismatic omphacite grains occur throughout the rock and also inside the large, poikiloblastic tourmaline grains. Further inclusions in tourmaline are glaucophane, rutile, titanite, albite, chlorite and epidote. Tourmaline grains are about 5 mm in diameter and 20–30 mm in length. Omphacite is homogeneous (Jd45Acm12Q43). The paragenesis glaucophane + chlorite + albite + omphacite + phengite + tourmaline, which is found in some places in the core of the eclogite block (represented by sample SY412) is the dominant mineral assemblage of sample SY413. Garnet is entirely transformed to chlorite, and rutile is replaced by titanite.

Sample SY420 is a schist from the matrix surrounding the blocks of the Syros mélange. The rock is foliated and folded and shows layering with alternating epidote- and chlorite-rich layers, both of which contain abundant tourmaline, phengite, garnet and albite. Accessories are allanite (in the core of epidote), rutile and titanite. Rutile forms large inclusions (500 µm) in garnet. Garnet has developed a chemical zonation with Mn-rich cores (Alm57Sps7Grs28Prp4Adr4) and Fe-rich mantles (Alm63Sps2Grs26Prp5Adr4). The mantle is partially resorbed and replaced by chlorite + albite. Some grains, however, show newly grown rims with Ca- and Mn-rich compositions (Alm26Sps34Grs31-Prp5Adr4). They show a sharp contrast to the Fe-rich mantle without diffusional equilibration. Tourmaline grains are 0·1–0·5 mm in diameter and 1–3 mm in length. Inclusions in tourmaline are epidote and rutile.

Sample SY441 is part of a blackwall that consists almost exclusively of coarse-grained chlorite and clusters of large tourmaline crystals (Fig. 3f), with only little omphacite and titanite and accessory allanite and rutile. Tourmaline crystals are about 5 mm in diameter and 30–50 mm in length, and show four zones in thin section. The cores (zone I) are light blue and sometimes contain inclusions of chlorite. The outer zones (II–IV) are free of inclusions and show greenish colours. Zone II is almost black and not developed in all grains. Zone III is dark green and the outermost zone IV is pale green.

Sample SY442 was taken from a blackwall wrapping an eclogite block with a thin black layer of tourmaline covering its surface (Fig. 3b). The blackwall is compositionally zoned with relatively sharp boundaries between the zones, changing with distance from the HP block. Sample SY442 was taken from a part consisting of omphacite and tourmaline crystals embedded in a matrix of albite (Fig. 3g and h). Tourmaline grains are 0·5–1·5 mm in diameter and 2–3 mm in length, and are concentrated in a layer of a few millimetres in thickness. They display four colour zones in thin section, with a light blue core (zone I), a green zone II, gradually fading into a dark, almost black zone III, which is sharply marked off from the green rim (zone IV; colours in plane-polarized light perpendicular to the c-axis; Fig. 3g). Omphacite inclusions are found in zones II–IV. Omphacite and albite (Ab99·2) in the rock matrix are closely intergrown and show petrographic equilibrium in thin section, suggested by sharp and euhedral grain boundaries (Fig. 3h). Omphacite is chemically zoned, with type-1 cores characterized by low jadeite contents (Jd36Acm22Q40) and type-2 rims with high jadeite contents (Jd54Acm19Q25).


    RESULTS
 TOP
 ABSTRACT
 INTRODUCTION
 ANALYTICAL METHODS
 GEOLOGICAL BACKGROUND
 INVESTIGATED SAMPLES
 RESULTS
 DISCUSSION
 SUMMARY AND CONCLUSIONS
 APPENDIX: EQUATIONS USED FOR...
 REFERENCES
 
PT conditions of formation of blackwalls and tourmaline
Temperature conditions during formation of tourmaline were calculated using the garnet–clinopyroxene Fe–Mg exchange thermometer, using the formulations by Ai (1994)Go and Krogh Ravna (2000)Go, and by using THERMOCALC (v 3.01; Powell & Holland, 1988Go; Powell et al., 1998Go). Only sample SY412 contains garnet in direct contact with type-2 omphacite, i.e. omphacite formed during the fluid influx that precipitated the tourmaline. Therefore, garnet–clinopyroxene thermometry was applied to garnet and type-2 omphacite in contact with one another in sample SY412 (Fig. 3e). The oxidation states of Fe in garnet and clinopyroxene were estimated by stoichiometric formula calculation. For an assumed pressure of 0·7 GPa, the thermometers of Ai (1994)Go and Krogh Ravna (2000)Go both give identical temperatures of 419°C. This is in agreement with the result from THERMOCALC (429°C). Accuracy of these calculated temperatures is not better than ±50°C, as a result of uncertainties in the determination of the oxidation state of Fe, and the activity models for clinopyroxene and garnet, as well as the uncertainties of electron microprobe analyses.

Determining the pressure conditions during formation of the reaction zones is hampered by the fact that the samples are silica-undersaturated and quartz is generally absent. The breakdown reaction of albite,

Formula 1(1)
has often been used as a barometer for the calculation of minimum pressures in eclogitic rocks containing omphacitic clinopyroxene and quartz. The paragenesis albite + omphacite in the absence of quartz provides maximum pressures of equilibration only. However, reaction (1) is not only pressure (and temperature) sensitive, but can also act as a buffer reaction for SiO2 during metasomatic processes. It has a positive slope in a P–log a(SiO2) diagram (Fig. 4). Addition of SiO2 by a fluid entering a rock containing omphacite + albite at constant pressure would produce more albite by reducing the jadeite component of the coexisting clinopyroxene. Fluids undersaturated in SiO2 with respect to the rock will remove SiO2 from the mineral assemblage by decomposing albite and increasing the jadeite component in omphacite, as for example:

Formula 2(2)
or

Formula 3(3)
Figure 4 shows the equilibria of albite + omphacite for two omphacites containing 33% and 50% jadeite, respectively. At constant temperature, reaction (2) could shift an assemblage from point A to point B by reducing a(SiO2) or proceed to point B' by increasing pressure. Omphacite in several samples (SY327, SY412, SY442) shows a two-stage zonation, with cores of low jadeite (Jd36–42) and rims of high jadeite (Jd52–54) contents. Assuming constant pressure, the first fluid had a higher silica activity, stabilizing Jd33 + Ab (point A in Fig. 4), whereas the second fluid had a significantly lower a(SiO2), producing Jd50 + Ab (point B in Fig. 4). The Ab + Omp equilibrium for a specific omphacite composition is shifted to higher pressures for higher SiO2 activities and reaches maximum pressures at a(SiO2) = 1 (i.e. in the presence of quartz). Therefore, a second, independent equilibrium is needed to determine pressure and silica activity during tourmaline formation.


Figure 4
View larger version (13K):
[in this window]
[in a new window]
[Download PowerPoint slide]
 
Fig. 4 P–log a(SiO2) diagram showing the upper pressure breakdown reaction of albite to omphacite + SiO2 in the SiO2-undersaturated field for pure albite and for omphacite with Jd33 and Jd50.

 
Samples SY309B, SY413 and SY412 show the mineral assemblage glaucophane + phengite + albite +chlorite + omphacite in equilibrium with tourmaline. The equilibrium of glaucophane, phengite, chlorite and albite or jadeite was calculated by the reactions


Formula 4(4)


Formula 5(5)

Figure 5a shows the calculated locations of reactions (1), (4) and (5) in a P–log a(SiO2) diagram, intersecting at the invariant point A at an assumed temperature of 400°C. Figure 5b displays the location of point A in the P–log a(SiO2) field for different temperatures and compositions. For a temperature of 400°C, point A is located at pressures of 0·61, 0·63 and 0·72 GPa for samples SY309B, SY413 and SY412, respectively. Silica activities in the three samples are 0·48 [SY309B; log a(SiO2) = –0·32], 0·42 [SY413; log a(SiO2) = –0·38] and 0·51 [SY412; log a(SiO2) = –0·29].


Figure 5
View larger version (12K):
[in this window]
[in a new window]
[Download PowerPoint slide]
 
Fig. 5 (a) Schematic P – log a(SiO2) diagram showing the intersection of reactions (1), (4) and (5) in invariant point A. The reactions Ab = Ne + SiO2 and Jd = Ne + SiO2 are also shown. (b) Location of invariant point A calculated for samples SY309B ({circ}), SY412 ({blacksquare}) and SY413 (•) in P – log a(SiO2) for different temperatures (given as numbers in °C).

 
In accordance with other workers (Trotet et al., 2001bGo; Rosenbaum et al., 2002Go; Putlitz et al., 2005Go) we conclude that the Syros rocks were exhumed along a near-isothermal decompression path; that is, showing a temperature decrease of ≤100°C during decompression to ~0·6 GPa. During exhumation of the unit, reaction zones formed at contacts between the contrasting lithologies, partly enhanced by mechanical mixing but also triggered by a strong influx of hydrous fluids, which reduced the activity of SiO2 in the rocks to values between 0·4 and 0·5. Reaction zones (blackwalls) are dominated by hydrous minerals such as chlorite, glaucophane, phengite, talc and Ca-amphibole, and contain abundant tourmaline that was precipitated during the formation of the blackwalls. PT conditions of tourmaline formation were 400–430°C and 0·62–0·72 GPa (Fig. 6). Therefore, a major influx of boron-rich hydrous fluid into the Syros mélange must have occurred during exhumation of the unit at a depth of 20–25 km.


Figure 6
View larger version (22K):
[in this window]
[in a new window]
[Download PowerPoint slide]
 
Fig. 6 PT diagram showing the prograde and retrograde PT path of Syros HP rocks. Peak conditions were taken from Okrusch & Bröcker (1990)Go and Trotet et al. (2001bGo). Conditions of tourmaline formation were determined during this study. The grey arrows mark retrograde PT paths for Syros eclogites and blueschists taken from Trotet et al. (2001bGo). The upper thermal stability limit of antigorite, experimentally determined by several workers (Ulmer & Trommsdorff, 1995Go; Wunder & Schreyer, 1997Go; Schmidt & Poli, 1998Go), was not reached in the rocks from Syros. P'90 is the PT path B for subducting slabs from Peacock (1990)Go. The continuous line is a 10°C/km geothermal gradient.

 
Chemical composition of blackwall tourmaline
Apart from XMg, variation of tourmaline chemical composition between the samples is very limited (Table 3). Formula calculations based on 31 oxygens and total Fe as Fe2+ result in 5·86–5·99 Si, 5·82–6·09 Al, 3·05–3·17 B and 2·93–3·25 (Mg + Fe + Ti) per formula unit. Syros blackwall tourmaline analysed in an earlier study showed a Fe3+/{Sigma}Fe ratio of ~0·30 (Marschall et al., 2004Go). TiO2 contents are between 0·12 and 1·14 wt % and vary between the zones of the crystals that were distinguished by colour in thin section. Ti contents are lowest in the light blue cores and higher in the dark green zones. The X-site is (almost) completely occupied by Na [0·92–0·98 cations per formula unit (c.p.f.u.)] and only minor Ca (0·01–0·05 c.p.f.u.). Analyses of F, Cl and H contents revealed no detectable Cl, very minor F (<0·03 p.f.u.) and high OH contents, >3·75 OH p.f.u., in most cases. Cr2O3 contents range from 0 to 0·28 wt %, which translates to ≤0·04 c.p.f.u. Cr. All other elements measured are of negligible concentrations in the tourmalines, including Mn, Zn, K and Li. Boron contents in excess of 3·00 c.p.f.u. in dravite, as well as high Na and OH contents, have been discussed by Marschall et al. (2004)Go. The calculated formulae are very close to the ideal schorl–dravite solid solution series and belong to the alkali-dravite group of Hawthorne & Henry (1999)Go with XMg between 0·58 and 0·82. Blue cores generally have higher XMg than green rims, with positive correlation of Fe, Ti and Ca contents.


View this table:
[in this window]
[in a new window]

 
Table 3 Chemical and B isotopic analyses of tourmaline

 
Boron isotopic composition of blackwall tourmaline
The small size of the SIMS spot during determination of B isotope ratios in tourmaline (≤5 µm) allows us to measure very detailed profiles even in small grains, and to unravel small-scale isotopic heterogeneities. During this study, traverses with 10–50 spots each were obtained for all eight samples.

The results of these measurements show four important features, as follows.

(1) 11B/10B isotopic ratios are very high, exceeding {delta}11B values of +18{per thousand} in all samples, with an extreme value of +28·4 ± 0·7{per thousand} (2{sigma}) in the tourmaline core of sample SY442 (Table 3, Fig. 7h).


Figure 7
View larger version (21K):
[in this window]
[in a new window]
[Download PowerPoint slide]
 
Fig. 7 Results of SIMS analyses of B isotopic composition of tourmaline from Syros. For each sample, at least one traverse of 10–50 single spot analyses was carried out across a grain. Average {delta}11B values are given by the numbers in the boxes. Error bars give analytical precision as 2RSDmean. (a) Sample SY309B; (b) sample SY327; (c) sample SY400; (d) sample SY412; (e) sample SY413; (f) sample SY420; (g) sample SY441. In sample SY441, no profile over one single grain was measured, but analyses from several grains (indicated by different symbols) were collected and sorted by location (zone I– IV) within the crystals. (h) Sample SY442.

 
(2) In seven out of eight samples intra-grain variation of {delta}11B is very limited, with profiles showing very little or a lack of internal zonation (Fig. 7). Differences between core and rim values are within analytical uncertainties (Table 3).

(3) Variation of {delta}11B between these seven different samples is also weak, ranging from +18·8 ± 0·9{per thousand} (sample SY441) to +22·2 ± 1·5{per thousand} (sample SY413).

(4) Sample SY442 is different from the other seven samples, as it displays a strong zonation in {delta}11B (Fig. 7h), which correlates with the optically visible zones (Fig. 3g). The blue core as well as zones II and III are extremely enriched in the heavy isotope, with {delta}11B values of ~+28{per thousand} (Table 3), whereas zone IV is significantly lighter. There is a sharp jump of ~6{per thousand} within 10 µm at the boundary between zones III and IV. Zone IV itself displays an internal isotopic zonation with {delta}11B increasing from +21·5{per thousand} in contact with zone III, to +24·0{per thousand} in the centre of zone IV, to ~+20·0{per thousand} at the edge of the tourmaline (Fig. 7h). A more detailed discussion of the {delta}11B profile in this sample is given below.


    DISCUSSION
 TOP
 ABSTRACT
 INTRODUCTION
 ANALYTICAL METHODS
 GEOLOGICAL BACKGROUND
 INVESTIGATED SAMPLES
 RESULTS
 DISCUSSION
 SUMMARY AND CONCLUSIONS
 APPENDIX: EQUATIONS USED FOR...
 REFERENCES
 
Boron concentration of blackwall-forming fluids
As no data on fluid inclusions related to the tourmaline-forming ‘event’ are available, the concentration of B in the fluids can be estimated only indirectly from B concentrations in minerals. Omphacitic clinopyroxene formed together with tourmaline from different samples within the blackwalls contains between 1·92 µg/g (SY309B) and 3·28 µg/g (SY441) B (Marschall, 2005Go). Combined with the clinopyroxene–fluid partition coefficient of 0·016 determined by Brenan et al. (1998b)Go, these abundances would imply B concentrations of the fluids in the range of 120–205 µg/g. Experimental studies by Morgan & London (1989)Go have shown that crystallization of tourmaline from solutions containing <0·2 wt % B2O3 (620 µg/g B) is possible, provided the fluids are low in pH (pH of quenched fluids <6·5). Morgan & London also demonstrated that tourmaline is not stable in alkaline fluids (pH >6·5) regardless of B concentration. Weisbrod et al. (1986)Go (cited by Dingwell et al., 2002Go) were able to crystallize dravite from Fe-free chlorite-bearing compositions in equilibrium with fluids containing between ~150 µg/g and ~300 µg/g B, in the temperature range of 350–450°C. This may indicate that the calculated B concentrations of ~100–200 µg/g are not too low for the formation of tourmaline.

Fractionation of boron isotopes
Theoretical calculations of isotope fractionation of boron predict preference of 10B for crystallographic sites tetrahedrally coordinated to oxygen, and 11B for trigonal sites. B in acidic hydrous fluids is predominantly trigonally coordinated in B(OH)3 units. Recent experiments have shown that this also holds true for high pressures up to 2·0 GPa (Schmidt et al., 2005Go). Boron in tourmaline is also typically trigonally coordinated with three BO3 groups per formula unit. Only small amounts of tetrahedral B were observed in tourmaline from Syros (Marschall et al., 2004Go), ranging from 0·1 to 0·2 B c.p.f.u. Therefore, at least 95% of the B in Syros tourmaline is trigonally coordinated. The tourmaline-forming hydrous fluid must have been acidic to stabilize tourmaline and, therefore, contained B in trigonal coordination [B(OH)3].

Experimental studies (e.g. Hervig et al., 2002Go and references therein; Heinrich et al., 2003Go; Wunder et al., 2004Go; Sanchez-Valle et al., 2005Go) have investigated the temperature dependence of B isotopic fractionation between phases of different B coordination (e.g. mica, amphibole, melt, fluid, tourmaline). Hervig et al. (2002)Go demonstrated a systematic decrease of fractionation with increasing temperature between phases containing tetrahedrally and trigonally coordinated boron, respectively, given by the formula


Formula 6(6)

where {alpha} is the fractionation factor and T is absolute temperature (K). For a temperature of 400°C, the assumed temperature for Syros blackwall formation, a fractionation of 12·6{per thousand} between phengite and fluid or phengite and tourmaline results from this equation. Applying the formulations of other researchers to this example results in very similar isotope fractionations of 12·6{per thousand} (Williams et al., 2001Go) and 12·0{per thousand} (Wunder et al., 2004Go). We applied the formulation of Hervig et al. (2002)Go, as it is based on the largest dataset, including natural and synthetic systems. The fractionation between tourmaline (0·15 BIV c.p.f.u.) and fluid would be 0·6{per thousand} applying equation (6), assuming that equilibrium fractionation is exclusively determined by the coordination of B in the phases. However, Palmer et al. (1992)Go have demonstrated experimentally a fractionation between tourmaline and fluid of ~4·5{per thousand} at 400°C and 0·2 GPa, which decreases with both increasing temperature and pressure.

Tourmaline from this study shows very homogeneous {delta}11B values even in large grains. As these grains were precipitated from hydrous fluid(s) entering the rocks, boron isotopic fractionation processes between the two phases can be monitored using these samples. The formation of tourmaline containing ~30 000 µg/g B significantly decreases the B concentration of any coexisting fluid. At low fluid/rock ratios, isotopic fractionation between tourmaline and fluid, as described by Palmer et al. (1992)Go, would lead to an increase of {delta}11B in the fluid over time and consequently to an isotopic zonation within the precipitated tourmaline grains, with {delta}11B values increasing from cores to rims. Flat {delta}11B patterns can be explained either (1) by a very high fluid/rock ratio, flushing the reaction zones with fluid of constant {delta}11B, or (2) by weak or absent B isotopic fractionation between tourmaline and hydrous fluid at the PT conditions of interest. Some of the investigated samples indeed show evidence for high fluid/rock ratios, as they contain large amounts of chlorite and no relics of their primary parageneses. Those samples are located at the outermost parts of the high-pressure blocks (SY441), or in the matrix (SY309B). However, SY412 was sampled from the core of an eclogite block that shows a gradual transition into blueschist and finally a chlorite-rich assemblage at its rim. SY412 still shows high proportions of its eclogitic paragenesis (Grt + Omp I + Rt), and only restricted hydration. This sample also contains large tourmaline grains with flat {delta}11B patterns (Fig. 7d). The petrographical observations suggest that (1) fluid/rock ratios strongly decreased from the matrix of the Syros mélange to the cores of the high-pressure blocks, and (2) fluid/rock ratios were rather low during retrogression of eclogite SY412, but significantly increased towards the block's rim (sample SY413). Fluid supply in the eclogitic core of the block was limited and the growing tourmaline must have had a strong impact on the B concentration of the fluid. Any isotopic fractionation between tourmaline and fluid would, therefore, have been documented in the precipitated tourmaline grains. The flat {delta}11B patterns in sample SY412 can, thus, only be explained by negligible B isotopic fractionation between tourmaline and hydrous fluid. This is in agreement with the fact that B in both tourmaline and acidic hydrous fluids is predominantly in trigonal coordination. In addition, Tonarini et al. (1998)Go reported {delta}11B values of tourmaline precipitated in miarolitic dykes on Elba island (Italy). Those workers suggested negligible B isotopic fractionation between tourmaline and fluid as a possible explanation for the absence of any significant B isotopic zonation within the tourmaline. However, these suggestions are in contrast to the experimental study of Palmer et al. (1992)Go at a pressure of 0·2 GPa.

B isotopic zonation in tourmaline of sample SY442
The strong B isotopic zonation in tourmaline of sample SY442 cannot be explained by fractionation effects between fluid and minerals during formation of the sample or by temperature effects. Instead, the zoning must be due to the precipitation of tourmaline from two different fluids of isotopically different composition entering the rock one after the other. This hypothesis is supported by the two-stage zonation of omphacite in this sample, showing the influx of a second fluid also leading to a drop in SiO2 activity. Omphacite in the matrix shows a two-stage zonation with cores of Jd36 and rims of Jd53. Omphacite included in tourmaline zones II and III (Fig. 3g) is relatively low in Na (Jd33), whereas the rims of omphacite included in tourmaline zone IV are high in Na (Jd45). Therefore, the two different omphacite generations can be related to different fluid pulses and growth stages of tourmaline, with a higher silica activity in the first fluid and a significantly lower a(SiO2) in the second fluid. Inclusions of type-1 omphacite in tourmaline zones I, II and III (high {delta}11B) and inclusions of type-2 omphacite occurring exclusively in zone IV, characterized by lower {delta}11B, support a genetic relationship between the two omphacite generations and the tourmaline generations. In some places, zone III is corroded and replaced by zone IV (Fig. 3g), suggesting that parts of the previously grown high-{delta}11B tourmaline were dissolved during influx of the lower-{delta}11B second fluid. This must have led to local mixing of B from the early tourmaline and the later fluid with their two contrasting isotopic compositions. The internal isotopic zonation of zone IV (Fig. 7h) is, therefore, best explained by mixing between B with {delta}11B ~ +28{per thousand} from early tourmaline and B with {delta}11B ~ +20{per thousand} from the second fluid. Hence, sample SY442 provides evidence for two fluids of different compositions.

Possible source of 11B-rich fluid
The formation of centimetre- to metre-sized reaction zones rich in OH-bearing minerals including tourmaline at 20–30 km depth during exhumation requires large amounts of H2O and boron. Within the mélange itself, there are no appreciable sources of H2O, because the metabasic rocks were already dehydrated on the prograde path, and because OH-bearing minerals in carbonates and schists, such as glaucophane and mica, survived the peak and retrograde metamorphism. Furthermore, the upper thermal stability of antigorite was not reached at Syros (Fig. 6). Thus, serpentinite, which is, in principle, an important carrier of H2O, did not dehydrate. In addition, the fact that the low {delta}11B values of minerals from the Syros high-pressure rocks were not significantly affected by late-stage fluid infiltration (Marschall, 2005Go) rules out these rocks as a boron source for tourmaline formation. Phengite ({delta}11B = –11·3 ± 1·4{per thousand}) and metamorphic tourmaline ({delta}11B = +0·9 ± 1·8{per thousand}) in a siliceous marble, antigorite ({delta}11B = +0·4 ± 7·8{per thousand}) from a serpentinite, tourmaline from an eclogite ({delta}11B = +2·8 ± 1·0{per thousand}), and metamorphic tourmaline from a metasedimentary schist ({delta}11B = –1·6 ± 1·1{per thousand}) from the Syros mélange (Marschall, 2005Go) all show that the fluids in equilibrium with these rocks would have had {delta}11B values between –3 and +3{per thousand} (marble, schist and eclogite) or about +13{per thousand} (acidic fluid in isotopic equilibrium with serpentinite at 400°C), significantly lower than that required for blackwall formation. Thus, it is concluded that H2O and boron were introduced into the mélange from external sources.

Modern subduction zone models involve a wedge or channel of exhuming rocks, which is located between the subducting and the overriding plate (Shreve & Cloos, 1986Go; Liou et al., 1997Go; Engi et al., 2001Go; Ernst, 2001Go; Guillot et al., 2001Go; Rubatto & Hermann, 2001Go; Gerya et al., 2002Go; Ring & Fleischmann, 2002Go; Roselle & Engi, 2002Go; Jolivet et al., 2003Go). Slices of HP rocks, sheared off from the subducting plate and tectonically mixed with serpentinites from the hanging wall, may rapidly ascend in these exhumation channels, while the subduction system is still active. Therefore, fluids released from the subducting plate will enter the overlying exhumation channel and rehydrate the HP rocks on their way back to the surface. The source of the B-rich fluids fluxing into the Syros mélange is probably also located within the plate, which continued to subduct during the exhumation of the Syros mélange (Fig. 8).


Figure 8
View larger version (27K):
[in this window]
[in a new window]
[Download PowerPoint slide]
 
Fig. 8 Sketch of vertical cross-section through a subduction zone showing the exhumation channel located between the subducting slab on the left and the overlying plate on the right. High-pressure rocks, now exposed on the island of Syros, were rehydrated during their exhumation by aqueous fluids, probably released from the progressively subducted slab.

 
Boron budget of fluids and rocks during dehydration
Major hosts of B in oceanic crust prior to subduction are low-temperature altered basalts and serpentinites that interacted with seawater at low temperatures, sediments (including pore water within the sediments) and altered basalts. B, therefore, is stored in the uppermost part of the subducting slab in the same reservoirs that contain most of the slab water. However, among these reservoirs, high {delta}11B values (>+5{per thousand}) are likely only in pore water, low-temperature altered basalts or serpentinites. B in altered basalts is mainly stored in clay minerals, either in their crystal structures or adsorbed on their surfaces (Harder, 1970Go; You et al., 1996bGo; Williams et al., 2001